Light confinement - Light Management - 利用埋入式次波長奈米結構提升非晶矽太陽能電池光捕捉效率

Ⅳ. Light Management

4.2 Light confinement

In contrast to anti-reflection coats, which increase the fraction of photons admitted to the cell, and concentration which increase the incident flux, light confinement techniques increase the path length of photons inside the cell, once admitted. Increasing the path length increases the probability of photogeneration per incident photon, particularly when the absorption coefficient is small, increasing the absorbed fraction. Light trapping is normally considered in the regime of geometrical optics where structures are large compared to the coherence length of the light, and light rays with different history do not interfere. This is a good approximation in silicon where cells are hundreds of microns thick. In micron scale structures, light should be treated as coherent and interference becomes important. In such systems, classical ray tracing approaches are not valid, and the photogeneration rate must be found from the gradient of the Poynting vector.

The simplest light trapping scheme is to introduce an optical mirror at the rear surface of the cell, either by metallising the rear cell surface or by growing the active layers on top of a Bragg stack. The mirror typically reflects over 95% of rays striking the rear surface. Rays which subsequently reach the front, semiconductor-air, surface are likely to pass through since the reflectivity of that interface must be small for efficient light capture. So the rear mirror effectively doubles the path length of the light. For an ideal mirror (with R = 1) and ideal front surface (with R = 0) the path length is 2w.

The simplest is where one surface is tilted relative to the other. Consider a rear surface tilted at an angle θ tilt relative to the planar front surface, as in Figure 4.4. When θ tilt ＞ 1/2*θ _c, normally incident rays will be reflected from the rear surface at an angle greater than θ c, and be totally reflected at the front. If a ray strikes the same portion of the rear surface on the second pass, it will be reflected at an even wider angle, and trapped again. For uniform cell width, both positive and negative tilt angles must be present, so that trapped rays will eventually be reflected at narrower angles and escape. If the positive and negative tilt angles are equal, then each ray makes a multiple of four passes across the cell.

Figure 4.4 Light trapping structure with a tilted rear surface, illustrating a ‗double bounce‘

light path. [6]

Ⅴ. Rigorous Coupled-Wave Analysis Method [22]

The diffraction of electromagnetic waves on periodic structures is an important problem with numerous physical and engineering applications. The core algorithm, which is based on Rigorous Coupled-Wave Analysis (RCWA), is a rigorous, fully-vectorial solution of Maxwell's equations [2-2][2-3]. The Rigorous Coupled-Wave Analysis (RCWA) method calculates the diffraction efficiency and field distribution for a 2D or 3D periodic structure. It helps in the design process of applications such as sub-wavelength structures, photonic band gap crystals, and other grating-assisted devices. The RCWA method splits the simulated structure into several parts with optical properties independent in the main propagation direction (i.e., the z direction in this study). Figure 5.1 shows a simple case with only one grating layer. The simulated region is defined by surperstrate, grating, and substrate region.

Figure 5.1 The schematic picture of a simple case in the RCWA method. The light is launched from the superstrate region. The analyzed structure is a grating. In the algorithm of RCWA method, all the periodic structures will be split up many such simple grating forms.

By factoring out an assumed time harmonic factor exp(-iωt) , Maxwell‘s equations can be expressed as: building blocks with a vertically homogenous region (i.e., a region with value of constant ε in the z direction). A complicated multilayer periodic structure can be decomposed into stacks of such basic building blocks. Then, the RCWA method applies the Fourier wave expansion in a homogeneous layer and Block wave expansion in a periodic structured layer, where has a variation of permittivity in the lateral direction. According to Bloch‘s Theorem, field components in a periodic layer can be expressed as:

(5-11)

(5-12)

(5-13)

(5-14)

The number of expanded waves is theoretically infinite. However, this is impractical in real numerical calculation. The wave expansions have a truncation called the Fourier harmonic number, which indicates the number of the expanded waves to solve the simulation.

Then, the |p| and |q| are the integers smaller than the Fourier harmonic number. The complex permittivity of the grating region is also defined by the summation of Fourier series. These waves are coupled to each other and the full vectorial Maxwell‘s equations to be solved in the Fourier domain. The diffraction efficiencies are then calculated at the end of simulation. The spatial field distribution with each Fourier harmonic number can be derived and then the total electromagnetic field is a sum over these fields. A detail derivation of the RCWA method and its open questions can refer to Ref. [23].

Ⅵ. Application of embedded sub-wavelength substrate on amorphous silicon thin film solar cell

Photovoltaic technology has received increased attention as one of the most promising approach to carbon-free energy production. In recent years, almost 90% of the photovoltaic industry market is dominated by traditional wafer-based silicon solar cells. But the cost of wafer-based silicon solar cells be limited by the cutting technology and is very difficult to reduce. Si-based thin-film solar cells have been a rising star in the photovoltaics industry in the last few years, owing to the advantages of less material usage and large area production [24]. Moreover, using nanostructures is also an appealing solution for increasing performances or lowering manufacturing cost of photovoltaic devices. However, the efficiency of thin film silicon solar cells critically depends on optical absorption in the silicon layer since silicon has low absorption coefficient in the red and near-infrared (IR) wavelength ranges due to its indirect bandgap nature. In typical thin film cells the thickness of the absorbing layer is governed by a tradeoff: the absorber must be optically thick to absorb a significat fraction of the incident photons at the same time the material has to be good enough to enable minority carrier collection lengths larger than the material thickness. These dual requirements largely define the cost per Watt of photovoltaic power. Therefore, for such thin-film structures, shortage of optical absorption length and insufficient optical coupling into the photo-active layer are the major challenges. Efficient light management of reducing the surface reflection as well as increasing the optical path for low energy photons are important, especially in the performance improvement and cost reduction.

In the past, a multilayer antireflection coating (ARC) was commonly used to reduce surface reflection [25][26]. On the other hand, light trapping also provide another method to increase the photon absorption, utilizing geometries to increase the optical path of the photons

near the band edge of the photo-active layer. Conventional light trapping scheme of thin film solar cells focuses on engineering the backside structures, which scatter or diffract the incompletely absorbed photons to oblique angles and hence enhance the propagation length within the absorption layers [27]-[29]. However, the backside structures only passively increase the optical path without the function of antireflection that actively introduce more photons into the cell. Therefore, geometries that simuntaneously reduces the broadband surface reflection and provides light trapping are highly desirable for high efficiency solar cells.

This section describes using the silicon nitride sub-wavelength structure on amorphous silicon solar cell to enhance the light flux, light trapping effect and power conversion efficiency. In traditional fabrication processes, first deposit an ITO layer as the front contact, and the thickness is about 80 nm. This thickness of the ITO layer is for the best anti-reflection effect of thin film at 600 nm wavelength. Sequentially it deposited the p, i, n layers of amorphous silicon on the ITO layer. Finally are another ITO layer as rear contact and an Al metal layer as back reflector. However, these each layer of tradition structure is constituted by flat films, and then it is undesirable of anti-reflection and light trapping effects. Therefore, we hope that a silicon nitride layer is added between the first ITO layer and glass. At the same time we make the sub-wavelength structure on the silicon nitride layer. This can maintain the shape of sub-wavelength structure on the following deposition films. We hope using a structure to meet the function of light trapping and anti-reflection effect. It is expected that this nanostructure can improve the short-circuit current density and power conversion efficiency of solar cells.

We combine both antireflection and light trapping mechanisms by using front pre-patterned nanostructures on the substrate to fabricate a-Si:H solar cells. The pre-patterned nanostructures were fabricated on the SiNx layer utilizing polystyrene (PS) colloidal

lithography, which is suitable for large area production. Then the following standard structures are reproduced layer by layer during the deposition of a-Si:H cells. We believe this novel patterning method is not restricted to a-Si:H solar cells, but is also widely applicable to other thin film materials.

6.1 Fabrication processes

The fabrication of our structure was done on a SiN_x layer deposited on a glass substrate using colloidal lithography followed by a reactive ion etching technique [30]-[33]. The choice of SiNx patterning could function as an index matching layer, while avoid potential damages resulting from patterning the frontal transparent conductive oxide. As illustrated in Figure 6.1(a), a 500 nm thick silicon nitride (SiN_x) layer was first deposited on a glass substrate via plasma-enhanced chemical vapor deposition (PECVD).

Figure 6.1 In this study, the process of pre-prepared samples. (a) The silicon nitride (SiN_x) of 500 nm was deposited on a glass substrate by using plasma-enhanced chemical vapor deposition (PECVD). (b)The polystyrene (PS) nanospheres was spun on the surface of the SiNx layer, forming a close-packed monolayer mask, and then shrink the size of the PS spheres.

Figure 6.2 The top view SEM images of the closely packed polystyrene nanospheres on the SiNx layer.

Then, polystyrene (PS) nanospheres with a plurality of 10 wt.% and diameter of 600 nm was spun-coated on the surface of the SiN_x layer, naturally arranging into a closely packed triangular lattice, which served as a self-assembled monolayer mask, and the SEM is shown in the Figure 6.2 [34]-[49]. Subsequently, inductively-coupled-plasma reactive ion etching (ICP-RIE) with incident oxygen (O) plasma was performed to shrink the size of PS spheres in order to facilitate the etching of SiN_x. The flow rate was set to be 20 sccm with a chamber pressure of 0.06 Pa and an RF power of 100 W. It is worth noting that the positions of PS spheres did not change during the size shrinking, as shown in Figure 6.1(b). The separations between PS spheres increased as their sizes decreased. Next, the ICP-RIE was performed on SiN_x and PS spheres with a CHF₃ /O₂ flow rate of 5/5 sccm, a chamber pressure of 1.33 Pa, and an RF power 150 W. The residual PS nanospheres were then removed by dipping into acetone with sonification for 5 min., as schematically shown in Figure 6.3(a). Figures 6.3(b) to 6.3(d) illustrate the deposition of a single-junction a-Si:H solar cell.

Figure 6.3 The device fabrication process of front pre-pattern substrate a-Si solar cell.

First, an 80 nm thick indium tin oxide (ITO) layer was deposited by DC sputtering. Then, the a-Si:H active layer with a total thickness of 292 nm (p/i/n=12/260/20 nm) was deposited using a high-density-plasma chemical-vapor-deposition (HDP-CVD) system with a growth temperate of 200^oC, a constant total pressure at 900 mTorr, and a plasma power density of 0.06 W/cm². Finally, an 80 nm-thick indium tin oxide (ITO) layer and 500 nm of aluminum were capped on top as the back electrode and reflector. Control cells were also fabricated on a flat glass substrate with a sputtered ITO layer and a commercially available Asahi U-type glass, denoted as the flat and the Asahi U cell, respectively. Figure 6.4(a) is the scanning electron microscopic (SEM) images of the resulting SiNx nipple nanopattern.

(a)

(d) (b)

(c)

Figure 6.4 Scanning electron microscopic (SEM) images of the fabricated SiNx nipple arrays:

(a) a 45-degree tilted top view and (b) a cross-sectional view.

The bottom width and height of nipple pattern can be controlled by varying the etching duration and conditions. As shown in Figure 6.4(b), the height and the bottom width of the patterns are ~450 nm and ~300 nm, respectively. Figure 6.5 shows a cross-sectional transmission electron microscopic (TEM) image of the fully fabricated solar cell on the pre- patterned substrate. The conformal deposition of individual layers is clearly resolved without cracks or voids that could adversely affect the device characteristics.

Figure 6.5 The cross-sectional TEM image of a fabricated solar cell on the pre-patterned substrate.

200nm

By the way, the commercial Asahi U-type substrate is a simple glass combined with one layer TCO. And the TCO layer is composed of SnO₂:F. However, this type of TCO layer will produce naturally variable degrees of surface roughness during the deposition. And the roughness R.M.S. is about 40nm. The extent of these surface roughnesses is random and not easy to control. The surface SEM image is shown in the Figure 6.6, and Figure 6.7 is the AFM image.

Figure 6.6 SEM image of the Asahi U-type sample

Figure 6.7 AFM image of the Asahi U-type sample

6.2 Experimental results and discussions

Our experiment structure is a embedded front pre-patterned nipple-shaped structure. This structure is mainly expected using the gradient refractive index effect (see Figure 6.6) to achieve anti-reflective and producing light trapping effect at the same time [50]-[56]. The light trapping effect is produced when the light touch to the nano-structure which bring the phenomenon of diffraction and scattering. This can change the travel direction of light and let the light have a longer optical path when it passes in the active layer. At the interface due to the relationship of incident angle is more prone to total reflection, and thus light can be confined to the active layer, to provide more opportunities for long-wavelength light absorption.

Figure 6.8 Schematic diagram of the sub-wavelength structure for antireflective by gradient refractive index. When the nano-structures is less than the wavelength of incident light, this anti-reflective nano-structure layer can use spatial gradient structure to achieve the effect of graded refractive index. The structure close to the end of air is a higher air ratio and a low refractive index, but the structure close to the end of structure is a lower air ratio and a high refractive index.

6.2.1 Integrating sphere reflectance measurement

To characterize the effects of antireflection and light trapping on these PPS solar cells, this shows the absorption spectrum measured (A=1-R) by an integrating sphere at normal incidence (see Figure 6.7). The cell that utilizes pre-patterned substrate shows superior absorption in the entire spectrum than that of the flat cell. The absorption enhancement is substantially increased in the wavelength range between 600 nm and 800 nm, resulted from the both light trapping and antireflection effects of the pre-patterned substrate. The PPS structure scatters the incident photons and enhances the transmittance of the substrate simutaneously. In addition, the enhanced absorption at the short wavelength range (λ<600nm), is resulted from antireflection effect by the effective refractive index only since the optical length for complete absorption is smaller than the thickness of p-i-n absorber layer.

Figure 6.9 The spectrum of cell absorption at normal incident.

6.2.2 External quantum efficiency measurement

Figure 6.8 is the external quantum efficiency (EQE) of three type structures. The EQE enhancement indicates the increased optical absorption. We also can see the enhancement cover most of the solar spectrum that is absorptive to a-Si:H solar cells. The same obvious phenomena of antireflection effect at short wavelength and light trapping at long wavelength also can be found in the EQE spectrum.

Figure 6.10 The external quantum efficiency (EQE) of the cell for three type substrates.

Figures 6.9(a) and 6.9(b) present the calculated enhancement factors of EQE of textured cells with respect to the flat control cell for wavelengths below and above 600 nm, respectively, where the enhancement factor is defined as Δ EQE=EQE/EQE_Flat. It can be seen in Figure 6.9(a) that the antireflection property of the PPSs increases the photocurrent

generation by roughly 20% for wavelengths between 400 nm and 575 nm, which is better than the Asahi U cell in this spectral range. For wavelengths longer than 575 nm, the photocurrent generation of the PPS cell achieves 2 to 12-fold enhancements, compared to the control cell, shown in Figure 6.9(b). The significant improvement in the infrared region indicates that the PPS can also effectively enhance the absorption by light trapping as well as the Asahi U.

Figure 6.11 The improvement factors of the EQE (Δ EQE) for EBN and Asahi U, (a) between 400 nm and 600 nm, and (b) between 600 nm and 800 nm.

Figure 6.12 is the internal quantum efficiency (IQE) of the cell for three type substrates.

Since we cannot directly measure the IQE of cells, so we obtained the result by through the calculation of EQE/absorption. From the figure, it can be seen that the three types of cells have almost the same IQE in the short wavelength. This situation shows that three components have almost the same growth characteristics of materials. But the flat cell curve has some ripple between the wavelength range of 575nm and 675nm. Speculated that the ripple is caused by the fluctuations of absorption curve, and this may be influenced by the constructive interference or destructive interference of light wave in the glass or SiNx layers.

Figure 6.12 The internal quantum efficiency (IQE) of the cell for three type substrates.

6.2.3 Power conversion efficiency measurement

The J-V measurement was performed under a simulated AM1.5G illumination condition (Oriel Class A 1000W) at room temperature following standard calibration and measurement procedures (see Figure 6.10) [57]. The detail electrical information is in the Table 6.1. The open voltage and fill factor for the PPS and flat cell remain aproximately the same. The increased efficiency (from 5.36% to 8.38%) is resulted from the enhanced short-circuit current density (from 12.89 mA/cm² to 19.77 mA/cm²). The PPS cell also shows superior conversion efficiency and current density to the cell that utilized with commerciallized Asahi U-type substrate.

Figure 6.13 J-V measurements of the flat reference, PPS, and Asahi U cells.

Table 6.1 The detail electrical information of J-V measurements.

6.3 Angular absorption

Furthmore, we measured the angle-resolved absorption spectroscopy of the reference flat cell and the PPS cell (see Figure 6.11). From the absorption color maps, it can be clearly seen that the PPS cell exhibits less dependency on wavelengths and incident angles than the flat cells. The PPS structure serves as an omnidirectional antireflective layer, which sufficiently couples the broadband oblique incident waves into the the absorber layer, up to 60°. The results guarantees sufficient light harvesting for the entire day.

Figure 6.14 The angle-resolved absorption spectroscopy for the cell with the (a) flat substrate and (b) PSS substrate.

6.4 Simulation

We also optimize the frontal pre-patterned substrate structures based on the theoretical calculation using a three-dimensional rigorous coupled wave analysis (RCWA) method. The approach has been utilized to investigate the diffraction and transmission properties of

nano-scale structures. In our simulation, we focus on the optimization of the polystyrene nanospheres of 600 nm period. So the simulated structural parameters include the bottom width and height of SiNx nipple pattern, as shown in Figure 6.12. In the implementation, the bottom width of SiNx nipple pattern can be controlled by the shrink size of the PS spheres before the etching process and the height of SiNx nipple pattern can be adjusted by the

在文檔中利用埋入式次波長奈米結構提升非晶矽太陽能電池光捕捉效率 (頁 47-0)